Posted 25 February 2007

E-mail discussion Printable version

Return to Discussion Text

Attendees/Participants
Alexander Arguello, Columbia University, New York
Nick Brandon, Wyeth Neuroscience-Princeton
Katherine Burdick, Zucker Hillside Hospital/AECOM
Miguel Camargo (via e-mail), Merck & Co.
Joe Callicott, National Institute of Mental Health, Bethesda, Maryland
Chris Carter, unaffiliated
Thomas Comery, Discovery Neuroscience, Wyeth Research, Princeton
Tom Fagan, Schizophrenia Research Forum
Jinbo Fan, Broad Institute, Cambridge, Massachusetts
Sydney Gary, Cold Spring Harbor Laboratory, New York
Hakon Heimer, Schizophrenia Research Forum
Barbara Lipska, NIMH, Bethesda, Maryland
Heather McKellar, Columbia University, New York
Huân Ngô, Yale University, New Haven, Connecticut
Misha Pletnikov, Johns Hopkins University, Baltimore, Maryland
David Porteous, University of Edinburgh, Scotland, UK
John Roder, Samuel Lunenfeld Research Institute, Toronto, Canada
Christopher Ross, Johns Hopkins University, Baltimore, Maryland
Ezra Susser, Columbia University, New York
Marquis Vawter, University of California, Irvine, California
Ryan Westphal, Bristol Myers Squibb Neuroscience, Wallingford, Connecticut

Note: The transcript has been edited for clarity and accuracy.

David Porteous
Hello, there. I have just come in from a long meeting, but ready to switch from gene therapy to schizophrenia and DISC1!

Chris Carter
Hello, everyone.

Hakon Heimer
Hello, Chris, and thanks for preparing the interaction figure. I bet you spent some late nights on that!

Chris Carter
Just a few, but well oiled.

Hakon Heimer
Hello, Tom and Misha.

Mikhail Pletnikov
Hello!

Tom Fagan
Hello, everyone.

Kate Burdick
Hi, Hakon and all.

Nicholas Brandon
Hi. It's Nick B here. Miguel Camargo sends his apologies that he is unable to join. He became a dad on Saturday!

Mikhail Pletnikov
Nick, Send him our congrats!

Hakon Heimer
I would like to introduce and thank our chat leader, David Porteous. In the informal spirit of this event, and because I think most people here are at least acquaintances, I won't go into any length about his biography and work, except to say that David is Head of the Medical Genetics Section at the University of Edinburgh, and that it was the work by this group that opened this promising footpaths into what one might call the Swamp of Schizophrenia Etiology. I'll now turn the floor over to David.

David Porteous
And as possibly the worst typist in the world, I will apologize now for my efforts on the keyboard, but thank you all for joining the DISC1 chat.

Tom Fagan
Don't worry, David, we don't care about typos. They will be edited out later....

David Porteous
Shall we get started then? First of all, great to see so many people in the chat room, and thanks for all the early comments. Could we start with a discussion about genetic evidence, heterogeneity and clinical phenotype? Christopher Ross had some comments on this, as did others. Thanks to all of those who have already posted several interesting comments summarized as: what insights can be gained for systems biology as to what DISC1 does? Can we predict phenotype from DISC1 allelotype? What are the strengths and weaknesses of modeling DISC1 function in the mouse? Recent DISC1 papers from Hennah, Taya, and Shinoda have also sparked comments. NB: these papers are nicely summarized by Tom Fagan on the Forum website. I note that there are comments that are based upon the assumption of a truncated protein being produced in the t(1;11) family. I have to remind you that there is no such evidence, but also agree that a dominant negative phenotype is possible, but by other mechanisms.

Let me say a couple of things. First, although there is now lots of evidence for DISC1 linkage and association, there is no proven mutation other than the t(1;11) translocation. Second, there are missense mutations that may be causative, but are low penetrant. And third, there is bound to be allelic heterogeneity. Any disagreement?

Mikhail Pletnikov
Dr. Porteous, I agree. It may be even more complicated because we are still uncertain about the translocation mutation, meaning if there is expression of mutant protein, and so on.

Christopher Ross
David, I agree. Barbara, I think your findings of decreases of DISC1 interactors are very interesting (see SRF related news story). Is there any sense that some patients have decreases in one set and other patients have decreases in another set? That might lead to subtyping. But it would probably require a very large sample.

Barbara Lipska
That’s not possible to test, even in our large cohort, but lymphoblast data may clarify some of the changes.

David Porteous
Kate, Barbara, and anyone from Helsinki, the data on DISC1 genetic “interaction” or “conditioned” effects on NDE1 and NDEL1 are provocative. Can we say anything more about this or the clinical correlations?

Kate Burdick
In our dataset the DISC Ser704Cys is not a risk locus, but it is the primary site of interaction with NDEL1 (haplotype); this may speak to some of the differential findings at this site.

Barbara Lipska
These "interacting" genes may also confer susceptibility on their own. We are in the process of testing this.

David Porteous
Misha, true, but following Barbara's point, what we have is good evidence for 50 percent expression in the t(1;11) lymphoblast cells (likewise for the phosphodiesterase 4 (PDE4) translocation for PDE4 activity) so I think this is one mechanism, and I would predict that downregulation of DISC1 (all isoforms) would be pathogenic.

Nick Brandon
Barbara, did you look at NDE1?

Kate Burdick
In our data NDEL1 does, in fact, confer risk on its own, but we have evidence of epistasis—in that the risk is only present in DISC1 704 Ser homozygotes.

Marquis Vawter
Barbara and Akira, are the same mRNA isoforms present in brain and lymphoblasts/lymphocytes?

Barbara Lipska
Mark Vawter, we're now trying to solve the puzzle of isoforms. There are some novel ones, and we cannot assay some already reported.

David Porteous
Marquis, yes, in our hands all DISC1 isoforms are detectable in LBCs, but we can’t conclude from that that there regulation is the same as in the brain.

Barbara Lipska
We are looking at NDE1. It appears weak so far.

Marquis Vawter
David and Barbara, I am not a DISC1er (yet); how many splice variants are there? Has anyone used the exon arrays with any luck on lymphocytes/brain for this question?

Barbara Lipska
In addition to L and Lv, we see at least four shorter ones (not fully confirmed yet)—all in brain.

David Porteous
DISC1, I think then, is the first case in the field where we can talk about epistasis. Can anyone think of another good example?

Barbara Lipska
COMT.

David Porteous
COMT episatic with? What is the current status?

Kate Burdick
Consistent with what Barbara is saying, we have similar data to theirs suggesting possible epistasis between COMT and DTNBP1. This COMT-DTNBP1 interaction is with cognition only—not disease.

Chris Carter
Regarding epistasis: see Nicodemus et al., 2007).

Barbara Lipska
David, see Nicodemus et al. from our group; it just came out.

Marquis Vawter
Barbara, thanks for the update on the isoforms.

David Porteous
DTNBP1 is linked by Miguel Camargo in his Y2H screens to DISC1 (see Camargo et al., 2007).

Barbara Lipska
David, and DTNBP1 links to microtubules.

David Porteous
So the systems biology approach is starting to give some insights and plausible new lines of enquiry.

Nick Brandon
The link in Miguel's work is through the plakin family.

Barbara Lipska
David, too many to handle!

Huan Ngo
David, epistasis is a very broad stroke. Can you folks specify more of what you are suggesting, evidence, and more importantly, suggestions about possible mechanisms?

Barbara Lipska
Huan, the paper above dealt only with statistical evidence.

David Porteous
Huan, you are correct, hence my parenthesis. It is going to be difficult to prove directly in human studies. Perhaps in the mouse it will be easier?

John Roder
Hi, David. Yes, if we had a good mouse model one could breed the two together and look for epistasis.

David Porteous
John, it is an important future route I think.

Mikhail Pletnikov
John, a few mouse models are coming soon, and we can test those questions using them.

Nick Brandon
John, what's a good mouse model?

Mikhail Pletnikov
Nick, I think it all depends on what question(s) you have.

John Roder
Yes, Mikhail, I believe so.

Christopher Ross
The question of mouse models raises again the issue, what does loss of DISC1 function look like? Has anyone reported a knockout? Whatever mouse models one has, I think a question would be to look at whether expression of DISC1 or its interactors would be decreased?

Mikhail Pletnikov
So we are coming to the point when we should be talking about a systems approach to this molecule to understand it. Does anyone have a good example from other candidate genes or mutations that we can draw on for some help? Chris, how about Huntington’s disease?

Christopher Ross
I would suggest that subtyping could come from the broad classes of interactors: alterations in proteins involved with cortical development might cause negative symptom-type schizophrenia; alterations in interactors involved with neuromodulation might cause a more affective phenotype. I will have to think about the huntingtin question.

David Porteous
Chris Carter has done the same sort of thing schizophrenia as he did for Alzheimer's disease. (Ed note: see Chris Carter comment).

David Porteous
I want to hear from anyone who has or knows of a DISC1 variant that acts like the t(1;11) as a dominant trait with reduced penetrance (you don't need to tell us what you have found, just if you have found something!).

Barbara Lipska
I would love to know what is "abnormal" about either mRNA or protein in schizophrenia or carriers of risk alleles! We found some novel (unconfirmed yet) isoforms whose expression appears associated with some high-risk polymorphisms (but again in subpopulations...), though they are not ready for prime time yet.

Marquis Vawter
It seems that the postmortem work that Barbara at NIMH is working on with isoforms would be a great place to start for developing mouse models based on human isoforms.

David Porteous
Chris, Nick, Misha, like clinical studies, the model is only as good as the phenotyping, but I think we can approach this as a top-down and bottom-up problem, that is, when we have a mouse that we think shows a phenotype of schizophrenia or bipolar affective disorder, is there evidence that the DISC1 pathway is perturbed, and if the DISC1 pathway is perturbed genetically, does the mouse have a schizophrenia or bipolar phenotype? A cautionary note would be to make sure we make clinically relevant mouse mutations.

Christopher Ross
I agree. Barbara, do you have some hypotheses for us that we could translate to mice?

Mikhail Pletnikov
Chris, I agree, but the mouse system can react differently, though I believe we still can address other pathways.

John Roder
Chris, with mice DISC1 might be loss or gain of function.

Christopher Ross
David, ah! And besides the translocation, which is hard to model, what is a clinically relevant mutation?

Hakon Heimer
David, all, what phenotypes are most attractive for testing future DISC1 mice? Working memory, prepulse inhibition—any others?

Barbara Lipska
Hakon, depending on your question, but it is complex, see our paper on modeling schizophrenia.

John Roder
Hakon, I believe working memory, sensory motor gating, depression, latent inhibition, and others…

Mikhail Pletnikov
I think that requiring a mouse model to mimic schizophrenia or bipolar disorder is too much. How about some endophenotypic features? These may be easily modeled in mice or in brain slices.

David Porteous
Chris, that must be the next stage of genetic studies—to go beyond association to biological tests for functional evidence and by large-scale genomic resequencing.

Christopher Ross
David, yes, I agree, but speaking as a mouse modeler now (with Misha), I think a key question is what mutations to make. Knockouts?

Mikhail Pletnikov
I agree with Barbara; different models will answer different questions. So, maybe more is better?

Hakon Heimer
David, given that the translocation t(1;11) co-segregated several major mental illnesses (seven schizophrenia, 10 major depression, one bipolar), must we think in terms of studying a gene that should really be called "disrupted in schizophrenia, major depression ..." and if so, does that complicate the picture enormously?

Barbara Lipska
Just remember that if we agree that schizophrenia is a polygenic disorder, by definition it will not be possible to model everything with one gene manipulation.

Tom Fagan
All, how about that spontaneous mutation in mice that truncates DISC1? How does that fit into the picture? Is it a useful model?

Mikhail Pletnikov
Tom, 129 DISC1 mutation does not look so simple as it may have in the beginning, but is a useful system, of course.

David Porteous
It’s a pity Jill Morris has left because she has started up a DISC1 program in zebra fish. What about other models to understand the basic biology of DISC1?

Christopher Ross
Yes I would like to hear more from Jill. I think zebra fish could be great. They don’t really have a cortex, but they do have a forebrain. What can one learn from animals much lower on the evolutionary scale? Genetic interactions?

David Porteous
Christopher Ross, and zebra fish can be used to look at peripheral nerves par excellence

John Roder
Hi, David. We have started a study in C. elegans as well.

Barbara Lipska
All, are we sure that this is "truncation," however defined?

David Porteous
Barbara, let’s give the genetic architecture question an airing. Studies designed to detect polygenic risk factors missed DISC1. It seems to me very unclear at this stage what proportion of the genetic liability for schizophrenia or bipolar is accountable by common variants and what proportion by rarer variants.

John Roder
Hi, Tom. Yes, I believe not only those exact DISC1 changes found in humans, but many others in this gene would be useful in establishing an allelic series.

Christopher Ross
I agree with John; phenotypes from a series of different mutants could be very interesting.

David Porteous
John, I agree. An allelic series will be key to this and may help sort out the genotype-phenotype question. Everyone, now that we have better and better information on the sites of protein interaction, we can speculate that some variants might have very specific effects (affecting, for example, splicing) and others have more general effects.

Mikhail Pletnikov
To all, I would like to find out more about non-neuronal DISC1? Does anyone have new data on this? Thanks.

David Porteous
Misha, it is high in the heart, positive in most tissues, including lymphoblast cells.

Mikhail Pletnikov
I guess I meant its expression in astrocytes, oligodendrocytes, microglia.

Barbara Lipska
We have done co-localization confocal studies showing very clearly that there is a lot of DISC1 in glial fibrillary acidic protein (GFAP)-positive cells.

Tom Fagan
Mikhail, do you think non-neuronal DISC1 may be as important as neuronal?

Mikhail Pletnikov
Barbara and Tom, yes. It is an interesting finding since it will help us to see how DISC1 can help us address gene-environment interactions in mouse and other models.

Marquis Vawter
The idea that DISC1 is mitochondrially localized (from David's work) has interesting implications for systems affected. However, briefly looking at the Camargo paper, I didn't recall mitochondrial proteins being found in the interactome.

David Porteous
Marquis, yes, a specific isoform 71 kDa is mitochondrial. Points up the strengths and weaknesses of a yeast two-hybrid screen—with the emphasis on “yeast”—that is, not human.

Nick Brandon
On interactome and mitochondrial proteins: It’s true we do not find too many mitochondrial proteins. I'd need to check with Miguel, but I think there were some weak interactors that may not have made the final edits.

Miguel Camargo
(via e-mail) I don't think the lack of mitochondrial proteins in the DISC1 interactome suggests a weakness in Y2H but rather a deficit in protein annotation. Annotation is incomplete for many proteins and we cannot assume at the moment that we know the localization of all the DISC1 interactors. DISC1 in itself can be found in different cellular compartments.

Marquis Vawter
David, the 71 kDa protein. Which mRNA is it: l, lv, s, es?

David Porteous
Marquis, I wish we knew for sure—it’s not trivial. Several groups are working on this. Barbara's for sure. Also Akira's and us (a bit).

Barbara Lipska
Yes.

Marquis Vawter
Barbara and David, we are screening some dorsolateral prefrontal cortex tissue right now for mtDNA gene expression, so I thought maybe a few key DISC1 mRNAs would be of interest. Any suggestions?

David Porteous
Can I ask for comments on the DISC1 pathway as a target for drug development?

Christopher Ross
David, well, PDE is an enzyme....

Barbara Lipska
As is Nudel.

David Porteous
Chris, and so is NUDEL (EORPA).

Christopher Ross
David, there are also drugs that act on microtubules.

David Porteous
All true. And PDEs are general targets, so we might get something back for efforts in other fields, but anti-PDE novel chemical entities often have side effects. Nick, any comments?

John Roder
Hi, David. One dream would be to isolate modifier genes in a genetic screen and use those as a new target for drug screens.

Christopher Ross
John, yes, that would be a good potential source for druggable targets. Any candidates?

John Roder
Not yet, Chris.

Nick Brandon
Drug targets to date: PDE4 is clearly an attractive target, and it’s clear that the Millar paper has influenced thinking in industry. NDEL1/EOPA is also very attractive but not validated. Beyond that, the interactome does open up other possible pathways to think about. We need to do the validation, though.

Chris Carter
Both DISC1 and GABAB receptors bind to ATF4. Could you use GABAB agonists/antagonists to study this area?

David Porteous
Nick, from an industry perspective, what will be most persuasive—genetic evidence, biological evidence, available mouse model, or others—for selecting targets?

Nick Brandon
David, all of the above! Ryan and others from industry, what do you think?

John Roder
Hi, Nicholas. What criteria are required for validation?

Ryan Westphal
I agree with Nick that the more validation information that connects a novel target to disease, the better.

Nick Brandon
John, very general, but you need to have some level of confidence that modulating your target is going to have a beneficial effect. As many of these proteins are novel, you need to understand their biology and do the experiments to understand the consequences of effecting them—mouse models, knockdown, tool compounds, etc.

David Porteous
We are 50 minutes into the discussion. Can I ask for burning questions not already raised, and can I pose one? What are the intermediate phenotypes we should be looking at?

Barbara Lipska
David, cognition.

Joseph Callicott
Our general experience with single candidate genes has been that we've seen larger effect sizes for functional and structural imaging phenotypes.

Kate Burdick
David, I am biased as a neuropsychologist, but the cognition data are fairly consistent now—early processing right up through verbal memory. An interesting thought is to also keep the intermediate phenotypes in mind when targeting drugs—they may be more amenable to treatment than the larger diagnostic phenotype.

David Porteous
Kate, I agree completely—thanks for that. It is a really important area that I think may be at the heart of the DISC1 effect.

Hakon Heimer
David, all, did we touch on whether there any studies in the works to look at associations between DISC1 genotypes against clinical phenotypes—deficit versus paranoid, etc.?

David Porteous
Hakon, you anticipated my question. Here, we are looking at brain imaging (as are Zucker and NIMH and others). The Finns and Ty Canon have good memory task data. We are also looking in non-schizophrenia or bipolar cohorts, for example, the Scottish Mental Study—cognitive aging.

John Roder
Yes, David, we need many more clinical endophenotypes, some of which will likely associate with gene x and others with gene y, and ideally not on drugs.

Joseph Callicott
David and Kate, I think one potential advantage of cognitive data over imaging is in the realm of subtyping. It has been more difficult studying large enough samples to subdivide.

Kate Burdick
Joseph, agreed. Our cognition dataset offers the power, but we, too, find larger effect sizes with imaging parameters.

David Porteous
Kate and Joe, and we can take cognitive tests beyond the clinically affected into first-degree relatives and the general population, the premise being that normal variation in DISC1 is likely to have an effect with clinical disorder.

Joseph Callicott
We've certainly taken that approach in the general population as a means of validating these SNPs of unknown functional impact, but our experience in first-degree relatives has been less satisfying, perhaps because these individuals have other genes or environmental factors that prevent fulminant illness.

Hakon Heimer
We have about 5 minutes in the official time, though the room is open for all those who want to "rush the speaker" or continue discussing. All will be included in the transcript.

David Porteous
Has anyone defined the DISC1 promoter?

Nick Brandon
Following on from David's last question—has anyone looked at the promoters of the interactors to relate back to Barbara's work?

David Porteous
With all the interaction data and the evidence for Barbara and others that these interactors are affected, perhaps we can subvert the usual criticism of mouse models (can't self-report on psychosis) and use biomarkers to validate?

Huan Ngo
David, Nicholas, more importantly, is there any evidence of methylation status and epigenetic control of the promoter, which would tie back to earlier discussion of environmental links and possibly epistasis?

David Porteous
[Post hoc reply] Nothing specific, but we have someone looking at this.

Hakon Heimer
David, all, are there any more kindreds identified with the translocation, or are there more family members in the original family?

David Porteous
[Post hoc reply] The translocation family has been under study semi-continuously for 30 years. Blackwood and colleagues (Blackwood et al., 2001) gave an update on the original publication of St. Clair and colleagues (St. Clair et al., 1990). It would be very unlikely to find other subjects with the same translocation that were not directly related. This is in all likelihood a relatively recent mutation. I recall a study presented as a poster at the WCPG in Quebec which looked in a Taiwanese cohort for evidence of the translocation, but as I would have predicted, found none.

Hakon Heimer
Our official time is up, so I'll thank David for preparing the discussion text and (in Irv Gottesman's terms) playing speed chess with multiple grand masters.

Mikhail Pletnikov
Thank you everyone and the organizers. It was a very interesting first-time experience!

Nicholas Brandon
Some interesting strands to follow up on. Cheers!

David Porteous
Congratulations to Miguel on becoming a dad! Hope all is well.

Hakon Heimer
Thanks to all the rest of you, and please stay in the room as long as you like

Kate Burdick
Very interesting discussion; thanks!!

David Porteous
All, I think Hakon is calling time up. It has been great to have such a lively discussion. Thanks, everyone.